From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mail-pl1-x634.google.com (mail-pl1-x634.google.com [IPv6:2607:f8b0:4864:20::634]) by sourceware.org (Postfix) with ESMTPS id 4D8D23858430 for ; Thu, 21 Apr 2022 23:50:15 +0000 (GMT) DMARC-Filter: OpenDMARC Filter v1.4.1 sourceware.org 4D8D23858430 Received: by mail-pl1-x634.google.com with SMTP id q3so7154601plg.3 for ; Thu, 21 Apr 2022 16:50:15 -0700 (PDT) X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20210112; h=x-gm-message-state:mime-version:references:in-reply-to:from:date :message-id:subject:to:cc; bh=cj2N6EagJnCslpevKXGlGpDg4HKlQKWCqOTLpRYm5g8=; b=r/QEQuTahvM6wmvctj5dtEkDOPwaOSqsF3FtoyYPIexXDZ3yUmIh0ccGYcnBDUJloW rWTirjEpJt/gpotX9Q5i9q8QZr/O3WSzJM+kqBk3zU/2/hA/xhnGnL+TJWDIgzxErBUj Fodytisg8Oig/ASE9zG5GRUBWjmXLFFLNTMbxb5tWE1d/wzuOdMDVzp/WC5vDeDLbsw1 REnDurdUtPTX9heE5uctebojWJ7qpy3P93TTtbZOmAeKQPknnNbbc1NP8F0nu0RTyY41 8AVdbmSlwqq+2N5GkN4QpzS0x62b6b5jROV4M0QbnzDTtSGnK6GUzktkr4G+BvG4dj2c aBBQ== X-Gm-Message-State: AOAM5335BBzYqTYnd5F54TSTlq6oVxp/iWP2qLaTFGl+BaI9rYtOBTdU dYhpPfkZ4aSlifn38mQe1Iiw4hDHW1R2nHe38kU= X-Google-Smtp-Source: ABdhPJy2zpt1RrONGfyCEgykX2TRYantjt1CBc8tJwC4ZGnpF/g9kGhHgU+9PMtIqOKlTXDbhlfICaYM6/F1s5zZEj8= X-Received: by 2002:a17:90b:4c8f:b0:1d2:90e4:26ef with SMTP id my15-20020a17090b4c8f00b001d290e426efmr12829538pjb.120.1650585014336; Thu, 21 Apr 2022 16:50:14 -0700 (PDT) MIME-Version: 1.0 References: <20220421031410.2142238-1-goldstein.w.n@gmail.com> In-Reply-To: From: "H.J. Lu" Date: Thu, 21 Apr 2022 16:49:38 -0700 Message-ID: Subject: Re: [PATCH v1 1/5] benchtests: Improve bench-strrchr To: Noah Goldstein Cc: GNU C Library , "Carlos O'Donell" Content-Type: text/plain; charset="UTF-8" X-Spam-Status: No, score=-3025.9 required=5.0 tests=BAYES_00, DKIM_SIGNED, DKIM_VALID, DKIM_VALID_AU, DKIM_VALID_EF, FREEMAIL_FROM, GIT_PATCH_0, RCVD_IN_DNSWL_NONE, SPF_HELO_NONE, SPF_PASS, TXREP autolearn=ham autolearn_force=no version=3.4.4 X-Spam-Checker-Version: SpamAssassin 3.4.4 (2020-01-24) on server2.sourceware.org X-BeenThere: libc-alpha@sourceware.org X-Mailman-Version: 2.1.29 Precedence: list List-Id: Libc-alpha mailing list List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Thu, 21 Apr 2022 23:50:17 -0000 On Thu, Apr 21, 2022 at 3:08 PM Noah Goldstein wrote: > > On Thu, Apr 21, 2022 at 3:13 PM H.J. Lu wrote: > > > > On Wed, Apr 20, 2022 at 8:14 PM Noah Goldstein wrote: > > > > > > 1. Use json-lib for printing results. > > > 2. Expose all parameters (before pos, seek_char, and max_char where > > > not printed). > > > 3. Add benchmarks that test multiple occurence of seek_char in the > > > string. > > > --- > > > benchtests/bench-strrchr.c | 126 ++++++++++++++++++++++++------------- > > > 1 file changed, 82 insertions(+), 44 deletions(-) > > > > > > diff --git a/benchtests/bench-strrchr.c b/benchtests/bench-strrchr.c > > > index abdae60c51..cceea77e1b 100644 > > > --- a/benchtests/bench-strrchr.c > > > +++ b/benchtests/bench-strrchr.c > > > @@ -23,6 +23,7 @@ > > > # define TEST_NAME "strrchr" > > > #endif > > > #include "bench-string.h" > > > +#include "json-lib.h" > > > > > > #define BIG_CHAR MAX_CHAR > > > > > > @@ -53,7 +54,8 @@ SIMPLE_STRRCHR (const CHAR *s, int c) > > > } > > > > > > static void > > > -do_one_test (impl_t *impl, const CHAR *s, int c, CHAR *exp_res) > > > +do_one_test (json_ctx_t *json_ctx, impl_t *impl, const CHAR *s, int c, > > > + CHAR *exp_res) > > > { > > > CHAR *res = CALL (impl, s, c); > > > size_t i, iters = INNER_LOOP_ITERS8; > > > @@ -61,8 +63,8 @@ do_one_test (impl_t *impl, const CHAR *s, int c, CHAR *exp_res) > > > > > > if (res != exp_res) > > > { > > > - error (0, 0, "Wrong result in function %s %p %p", impl->name, > > > - res, exp_res); > > > + error (0, 0, "Wrong result in function %s %p %p", impl->name, res, > > > + exp_res); > > > > These changes aren't needed. > > > > > ret = 1; > > > return; > > > } > > > @@ -72,24 +74,28 @@ do_one_test (impl_t *impl, const CHAR *s, int c, CHAR *exp_res) > > > { > > > CALL (impl, s, c); > > > } > > > - TIMING_NOW (stop); > > > > > > + TIMING_NOW (stop); > > > > Not needed. > > Will fix in V2 > > > > > TIMING_DIFF (cur, start, stop); > > > > > > - TIMING_PRINT_MEAN ((double) cur, (double) iters); > > > + json_element_double (json_ctx, (double) cur / (double) iters); > > > + return; > > > > Return isn't needed. > > Will fix in V2. > > > > > } > > > > > > static void > > > -do_test (size_t align, size_t pos, size_t len, int seek_char, int max_char) > > > +do_test (json_ctx_t *json_ctx, size_t align, size_t pos, size_t len, > > > + int seek_char, int max_char, size_t freq) > > > /* For wcsrchr: align here means align not in bytes, > > > but in wchar_ts, in bytes it will equal to align * (sizeof (wchar_t)) > > > len for wcschr here isn't in bytes but it's number of wchar_t symbols. */ > > > { > > > size_t i; > > > + size_t pos_chunk_sz = freq ? (pos / freq) : pos; > > > + size_t last_pos = len; > > > CHAR *result; > > > CHAR *buf = (CHAR *) buf1; > > > > > > - align &= 7; > > > + align &= (getpagesize () - 1); > > > > If we have such large alignments, the tests may be skipped. > > Should we change it to 127 instead? > > There is logic around page cross cases in x86_64 versions so think > makes sense to support benchmarking it. > > Also i think that would tend to give the previous version a bit of > an unfair disadvantage as the slow aligning case will never be > tested in the new version. If "align" is close to the page size, will it trigger if ((align + len) * sizeof (CHAR) >= page_size) return; and skip page cross cases? > > > > > if ((align + len) * sizeof (CHAR) >= page_size) > > > return; > > > > > > @@ -103,6 +109,16 @@ do_test (size_t align, size_t pos, size_t len, int seek_char, int max_char) > > > if ((i > pos || pos >= len) && buf[align + i] == seek_char) > > > buf[align + i] = seek_char + 10 + (random () & 15); > > > } > > > + > > > + if (pos_chunk_sz == 0 && pos) > > > + pos_chunk_sz = 1; > > > + > > > + for (i = pos_chunk_sz; i < pos && i < len; i += pos_chunk_sz) > > > + { > > > + buf[align + i] = seek_char; > > > + last_pos = i; > > > + } > > > + > > > buf[align + len] = 0; > > > > > > if (pos < len) > > > @@ -110,66 +126,88 @@ do_test (size_t align, size_t pos, size_t len, int seek_char, int max_char) > > > buf[align + pos] = seek_char; > > > result = (CHAR *) (buf + align + pos); > > > } > > > + else if (last_pos < len) > > > + result = (CHAR *) (buf + align + last_pos); > > > else if (seek_char == 0) > > > result = (CHAR *) (buf + align + len); > > > else > > > result = NULL; > > > > > > - printf ("Length %4zd, alignment in bytes %2zd:", len, align * sizeof (CHAR)); > > > + json_element_object_begin (json_ctx); > > > + json_attr_uint (json_ctx, "len", len); > > > + json_attr_uint (json_ctx, "pos", pos); > > > + json_attr_uint (json_ctx, "align", align); > > > + json_attr_uint (json_ctx, "freq", freq); > > > + json_attr_uint (json_ctx, "seek", seek_char); > > > + json_attr_uint (json_ctx, "max_char", max_char); > > > + json_array_begin (json_ctx, "timings"); > > > > > > FOR_EACH_IMPL (impl, 0) > > > - do_one_test (impl, (CHAR *) (buf + align), seek_char, result); > > > + do_one_test (json_ctx, impl, (CHAR *) (buf + align), seek_char, result); > > > > > > - putchar ('\n'); > > > + json_array_end (json_ctx); > > > + json_element_object_end (json_ctx); > > > } > > > > > > int > > > test_main (void) > > > { > > > - size_t i; > > > + json_ctx_t json_ctx; > > > + size_t i, j; > > > + int seek; > > > > > > test_init (); > > > + json_init (&json_ctx, 0, stdout); > > > > > > - printf ("%20s", ""); > > > - FOR_EACH_IMPL (impl, 0) > > > - printf ("\t%s", impl->name); > > > - putchar ('\n'); > > > + json_document_begin (&json_ctx); > > > + json_attr_string (&json_ctx, "timing_type", TIMING_TYPE); > > > > > > - for (i = 1; i < 8; ++i) > > > - { > > > - do_test (0, 16 << i, 2048, 23, SMALL_CHAR); > > > - do_test (i, 16 << i, 2048, 23, SMALL_CHAR); > > > - } > > > + json_attr_object_begin (&json_ctx, "functions"); > > > + json_attr_object_begin (&json_ctx, TEST_NAME); > > > + json_attr_string (&json_ctx, "bench-variant", ""); > > > > > > - for (i = 1; i < 8; ++i) > > > - { > > > - do_test (i, 64, 256, 23, SMALL_CHAR); > > > - do_test (i, 64, 256, 23, BIG_CHAR); > > > - } > > > - > > > - for (i = 0; i < 32; ++i) > > > - { > > > - do_test (0, i, i + 1, 23, SMALL_CHAR); > > > - do_test (0, i, i + 1, 23, BIG_CHAR); > > > - } > > > + json_array_begin (&json_ctx, "ifuncs"); > > > + FOR_EACH_IMPL (impl, 0) > > > + json_element_string (&json_ctx, impl->name); > > > + json_array_end (&json_ctx); > > > > > > - for (i = 1; i < 8; ++i) > > > - { > > > - do_test (0, 16 << i, 2048, 0, SMALL_CHAR); > > > - do_test (i, 16 << i, 2048, 0, SMALL_CHAR); > > > - } > > > + json_array_begin (&json_ctx, "results"); > > > > > > - for (i = 1; i < 8; ++i) > > > + for (seek = 0; seek <= 23; seek += 23) > > > { > > > - do_test (i, 64, 256, 0, SMALL_CHAR); > > > - do_test (i, 64, 256, 0, BIG_CHAR); > > > + for (j = 1; j < 32; j += j) > > > + { > > > + for (i = 1; i < 9; ++i) > > > + { > > > + do_test (&json_ctx, 0, 16 << i, 2048, seek, SMALL_CHAR, j); > > > + do_test (&json_ctx, i, 16 << i, 2048, seek, SMALL_CHAR, j); > > > + } > > > + > > > + for (i = 1; i < 8; ++i) > > > + { > > > + do_test (&json_ctx, i, 64, 256, seek, SMALL_CHAR, j); > > > + do_test (&json_ctx, i, 64, 256, seek, BIG_CHAR, j); > > > + > > > + do_test (&json_ctx, i * 15, 64, 256, seek, SMALL_CHAR, j); > > > + do_test (&json_ctx, i * 15, 64, 256, seek, BIG_CHAR, j); > > > + } > > > + > > > + for (i = 0; i < 32; ++i) > > > + { > > > + do_test (&json_ctx, 0, i, i + 1, seek, SMALL_CHAR, j); > > > + do_test (&json_ctx, 0, i, i + 1, seek, BIG_CHAR, j); > > > + } > > > + if (seek == 0) > > > + { > > > + break; > > > + } > > > + } > > > } > > > > > > - for (i = 0; i < 32; ++i) > > > - { > > > - do_test (0, i, i + 1, 0, SMALL_CHAR); > > > - do_test (0, i, i + 1, 0, BIG_CHAR); > > > - } > > > + json_array_end (&json_ctx); > > > + json_attr_object_end (&json_ctx); > > > + json_attr_object_end (&json_ctx); > > > + json_document_end (&json_ctx); > > > > > > return ret; > > > } > > > -- > > > 2.25.1 > > > > > > > > > -- > > H.J. -- H.J.